This current research project was conducted at The University of the South Pacific (USP), a regional higher education provider in the Pacific Oceanic region. The university is jointly operated by 12 member nations and services an area of 30 million square kilometers across the Pacific Ocean, with more than 30,000 students enrolled in total. Ongoing studies continue to provide deeper insight into the learning styles and patterns of Pacific students, with courses and content delivery being modified accordingly (Sharma et al., 2018, 2020). For example, educators often incorporate the “Pacific way of learning” into their courses, which is any informal education that does not include classroom-based or curriculum-based learning. The uniqueness of Pacific Oceanic geography and multi-national governance structure of USP provides a novel student experience which is currently understudied. The continued diaspora of Pacific peoples away from the region, particularly those obtaining higher education in the region’s Universities, also necessitates more investigation into undergraduate student experience.

There have been previous studies on student attitude toward science (including several studies looking at chemistry specifically) in this regional population, for example, Naiker et al. (2020a) used Test Of Science Related Attitudes (TOSRA) to evaluate the attitude of secondary Fijian students toward science, finding a generally positive attitude, although differences were observed based on gender and ethnicity. Also, Sharma et al. (2021) found slightly poorer attitudes among secondary Tongan students, although there was no difference between genders in this population. Others (Brown et al., 2014a) used the Attitude to the Study of Chemistry Inventory (ASCI) instrument to quantify attitude toward chemistry in cohorts of first-year undergraduate and foundation chemistry students. Students’ attitudes were generally positive, with Brown et al. (2014a) identifying three constructs of attitude measured by the instrument (affective, cognitive and value). However, no correlation between attitude and achievement was conducted.

Previous studies have reported a moderately positive correlation between student attitude and their achievement in university-level courses (Osborne et al., 2003; Xu and Lewis, 2011; Xu et al., 2012). However, data correlating the attitude and achievement of students studying chemistry is quite limited (Bauer, 2008). Among the major branches of science, chemistry stands as a unique topic, generally requiring more theoretical principles and mathematical knowledge compared to biology, but also more hands-on laboratory skills compared to fields such as physics. Hence there may not be a strong correlation between student attitude and their achievement in chemistry, as other internal factors influencing student achievement, such as mathematical knowledge, also affect student achievement to a considerable degree. Indeed, previous studies looking specifically at the relationship between attitude and achievement in chemistry students have only found a weak positive correlation (Bauer, 2008; Xu and Lewis, 2011; Brown et al., 2015). Hence in this study we sought to clarify the relationship between student attitude and achievement amongst undergraduate chemistry major students at a regional university in the South Pacific region.

Difficulties arise when investigating student attitude toward a given topic, as the concept of “attitude” is not well understood and remains a “somewhat nebulous” idea (Brown et al., 2014a). Part of the confusion stems from the fact that attitude is not a unidimensional construct (Reid, 2006). Rosenberg et al. (1960) proposed a tripartite model of attitude, comprising cognitive, affective, and behavioral elements. In this model, the cognitive component encompasses the student’s knowledge and belief surrounding the topic in question; the affective component includes their individual preferences about the topic; while the behavior component incorporates how the student tends to act regarding the topic. Bagozzi and Burnkrant (1979) later proposed a two-component model of attitude, comprising only the cognitive and affective elements, arguing that student behavior was largely derived from their cognitive and affective attitude toward the topic, rather than being a distinct entity of student attitude. In other words, student attitude toward the subject of chemistry could be thought of as how the students think (cognitive element) and feel (affective element) toward chemistry.

There is a growing body of research on student attitude, achievement, and the relationship between these two variables. Work by Majid and Rohaeti (2018) demonstrated that a positive attitude toward learning improved motivation and learning outcomes for students. Similarly, Kahu and Nelson (2018) concluded that student attitude is a critical factor that determines student achievement. They argue that although other factors are related to student achievement, students are responsible for their learning and play a central part in the teaching and learning process.

The attitude of students toward science is the determining factor of emotions such as “liking” or “disliking” science subjects, affecting their learning outcomes and performance (Kurniawan et al., 2019). Such attitudinal constructs can either guide students to study the subjects in related fields or subjects they like, or they can push students away from careers that require even moderate competencies in the subject (Recber et al., 2018). There is generally a positive attitude toward science amongst secondary students; however, a smaller number of students still have a negative attitude (Naiker et al., 2020a). In such cases, teacher motivation and learning design play important roles in garnering student attention and improving students’ attitude toward science subjects. Although there has been a sustained level of research into general attitudes toward science using instruments such as the Science Attitude Scale (SAS; Bajaj and Devi, 2021) and the TOSRA (Sharma et al., 2021), there is a recent trend to consider more specific science disciplines, such as chemistry (Montes et al., 2018; Rüschenpöhler and Markic, 2020). This follows from studies demonstrating a visible decrease in student interest and achievement in specific science disciplines (Vilia et al., 2017) and subsequent impaired progress to related undergraduate programs in Higher Education institutes (Sharma et al., 2018). Studies on the effect of student attitude toward achievement have been conducted across a range of subjects such as mathematics, English, physics, and chemistry (Vilia et al., 2017; Dimosthenous et al., 2020; Kaur and Vadhera, 2021; Sharma et al., 2021). However, we focus specifically on studies from the chemistry discipline in the remainder of this section.

Studies investigating student achievement and success in chemistry have demonstrated negative impacts of various issues including unappealing teaching methods (Najid et al., 2021), lack of enjoyment in the subject (Majid and Rohaeti, 2018), and disinterest of students toward the subject (Duangsri et al., 2017). However, there is evidence that positive student attitudes toward the subject of chemistry can correlate with student achievement (Brown et al., 2015; Vilia et al., 2017; Kousa et al., 2018; Montes et al., 2018; Kenni, 2019; Wahyudiati and Rohaeti, 2020). Other work by Fraser and Lee (2015) has indicated that student attitude is partially responsible for student achievement, while other factors such as self-efficacy, motivation and the learning environment were also significant contributors. To explore the potential impact of these additional factors, a range of studies have been conducted to evaluate the attitude of students and/or their achievement in chemistry from different perspectives and dimensions such as gender-based differences (Steegh et al., 2021), the influence of teacher practices (Mwangi et al., 2020), the self-efficacy of students (Kadioglu-Akbulut and Uzuntiryaki-Kondakci, 2021) and their experiences in learning chemistry (Kaur and Vadhera, 2021; Naiker et al., 2021). Most notably, Celik (2018) showed that providing students with an interactive and engaging learning environment only enhance their achievement if they had a positive attitude toward the topic, otherwise no improvement in achievement was observed. This highlights the importance of Higher Education providers assessing and understanding the attitude toward chemistry amongst their student cohorts.

The current research will attempt to identify any possible relations between student attitude and achievement in undergraduate introductory chemistry at USP. The internal structure of the instrument used to quantify attitude will also be explored (factor analysis), and a grouping technique (cluster analysis) will be applied to student achievement scores to identify different levels of academic ability. Relations between attitude and academic performance will then be explored in any groups identified, using correlation.

For the assessment of student attitudes toward the study of chemistry, this study utilized the ASCI instrument, as developed by Bauer (2008). Several instruments have been developed specifically for measuring student attitude and experiences in the study of chemistry, including the Chemistry Attitudes and Experiences Questionnaire (CAEQ; Wahyudiati and Rohaeti, 2020; Naiker et al., 2021), the ASCI (Bauer, 2008), its refined version, the ASCIv2 (Xu et al., 2012), amongst others (Salta and Tzougraki, 2004; Cheung, 2009). The ASCI and ASCIv2 have been used by numerous researchers for quantifying student attitudes toward chemistry (Xu and Lewis, 2011; Brown et al., 2014a,2015; Kahveci, 2015; Mooring et al., 2016; Sen et al., 2016; Sen and Oskay, 2017), making them among the most widely used and validated instruments for this purpose. Both instruments display two distinct subscales consistent with the two elements of attitude proposed by Bagozzi and Burnkrant (1979) (i.e., cognitive and affective). The instrument utilizes a semantic differential response, with students asked to position their response on a 7-point scale (between two opposing adjectives, e.g., good-bad, valuable-worthless). The use of this approach avoids students “reading” social desirability or directionality into the item, hence maximizing the focus of the instrument upon students’ individual attitudes toward chemistry. To minimize response bias, the positive adjective is presented first for some of the items, while the negative adjective is presented first for the remaining items.

Ethical approval for this study was obtained from the Human Research Ethics Committee of the host university (The University of the South Pacific). The study population was the 2013 cohort of undergraduate chemistry major students at USP who were enrolled in a first-year chemistry unit (CH102; Principles and Reactions of Organic Chemistry). This unit is compulsory for students enrolled in the Bachelor of Science degree, with either a major or minor in chemistry (Wakeling et al., 2014). All students enrolled in this course must have previously passed chemistry in high school or completed a foundational chemistry unit at university. The unit content was delivered in a face-to-face format via standard 4 × 50-min lectures per week throughout the 14-week semester and included a weekly 1 h tutorial session and 12 × 3 h practical laboratory classes throughout the semester. It was also supported by teaching material and other resources uploaded to the online learning platform (Moodle). Half of the overall grade for the unit is derived from student performance on continuous assessment items throughout the term (two short tests and laboratory-based assessment), while the final exam provides the other half of their grade. The laboratory assessment component included lab reports, as well as a final lab skills-based assessment. To pass the unit, students must achieve at least 40% on the final exam and 50% overall for the unit, as well as at least 60% tutorial attendance and 75% laboratory attendance (Gopalan, 2014).

To measure student achievement in the unit, the scores for each student on the continuous assessment items and final exam were utilized. To obtain the overall grade for the unit, these scores were summed. Grades data were available for a total of 235 students. In addition, the attendance rates of individual students at the laboratory and tutorial classes were also recorded. It should be noted that marks for individual assignments were not available for all students; hence only the results of 221 students are included in analysis relating to grades achieved on individual assessment items. All students were invited to complete the ASCI instrument, with 97 students opting to provide informed consent and allow their grade data to be anonymously linked to their ASCI survey responses. A total of 95 valid responses were provided; only these responses were used in subsequent analyses correlating student attitudes with their grades. The cognitive sub-scale for each student was calculated by summing the scores for questions 1, 4, 5, and 10, while the affective sub-scale was calculated by summing responses to questions 7, 11, 14, and 17; following the methods of Brown et al. (2015). For all items, a higher rating indicates a more positive attitude toward chemistry. Exploratory factor analysis using principal component analysis with varimax rotation was performed in SPSS (v26), to confirm if the subscales identified in the present dataset matched those subscales identified by previous researchers. All other statistical analyses were performed in R Studio, running R 4.0.2 (R Core Team, 2020). Cluster analysis using the K-means method was performed on students’ academic achievement scores. K-means clustering is defined as a method having an iterative process in which the dataset is grouped into a number (K) of pre-defined non-overlapping clusters, making the inner points of the cluster as similar as possible while trying to keep the clusters at distinct space. The method allocates the data points to a cluster so that the sum of the squared distance between the clusters centroid and the data point is at a minimum, at this position the centroid of the cluster is the arithmetic mean of the data points that are in the clusters (Rosenberg et al., 2020).

The average attendance throughout the semester was 89.1% for the laboratory classes and 81.6% for the tutorials. Attendance at lecture sessions was not monitored. The average overall mark for the unit was 49.7%, with most students achieving a passing grade (Figure 1).

Initially, exploratory factor analysis (principal component analysis with varimax rotation) was conducted on the 95 responses to the ASCI instrument. The Kaiser-Meyer-Olkin measure was 0.790, indicating an adequate sample size for this analysis, while Bartlett’s test of sphericity was P < 0.001, indicating that responses to the 20 variables (statements) were inter-related. Examination of the scree plot from the initial analysis suggested an optimum solution of three factors with eigenvalues >1, which together accounted for 45.6% of the variance. The exploratory factor analysis was then optimized for a three-factor solution, which provided the factor loadings shown in Table 1 (explaining 51.3% of variance). These differed slightly from previous research on similar student populations (Brown et al., 2014a), principally for item numbers 1, 8, 13, and 19 (Table 1). The majority of items loaded onto the first factor, which appeared to be predominantly related to student interest and the level of mental challenge (e.g., Interesting – Dull, Comfortable – Uncomfortable, Satisfying – Frustrating). The second factor contained items relating to workload management and subject worth (e.g., Chaotic – Organized, Tense – Relaxed, Worthless – Beneficial). All three items loading onto the final factor were related to topic clarity and understanding (Confusing – Clear, Complicated – Simple, Easy – Hard).

To investigate any natural “grouping” or “clustering” of students with differing achievement levels, cluster analysis was performed on the students’ continuous assessment, final exam and overall scores. This revealed the presence of a low-achieving (LA) cluster comprising 38% of the cohort, with a high-achieving (HA) cluster comprising the remaining 62% of students, with the cluster ratio being 1:1.64.

As shown in Table 2, the LA cluster did not show any significantly lower attendance rates in the tutorial classes but did have significantly lower marks on the continuous assessment items and final exam, leading to lower overall grades for the unit (P < 0.001). In addition, the attendance at laboratory classes was much lower in the LA cluster (P < 0.001). However, although the LA cluster had slightly lower scores on both the cognitive and affective components of the survey instrument, this difference was not significant (P > 0.05).

Among the entire student cohort, the rate of tutorial attendance was positively correlated with the marks achieved on the continuous assessment items (r₉₅ = 0.152, P < 0.05) but negatively correlated with the affective component of attitude (r₉₅ = −0.219, P < 0.05). Attendance at laboratory classes was strongly positively correlated with the marks on the continuous assessment items (r₉₅ = 0.594, P < 0.01), the final exam (r₉₅ = 0.388, P < 0.01), and the overall grade for the unit (r₉₅ = 0.510, P < 0.01). There was a positive correlation between the cognitive and affective components of attitude among students in the LA cluster (r₃₅ = 0.412, P < 0.05), but not in the HA cluster (r₆₀ = 0.105, P > 0.05).

As shown in Table 3, the affective component of student attitude toward chemistry was positively correlated with all assessment items (continuous assessment, final exam, and overall grade). However, it was negatively correlated with tutorial attendance. In contrast, the cognitive component was not significantly correlated with any assessment item or class attendance. In other words, the way students “felt” about the subject took precedence over what they “thought” about it.

For students in the LA cluster, the affective component was significantly correlated with their score on the final exam (r₃₅ = 0.512, P < 0.01) but not the continuous assessment items. In the HA cluster, the affective component was positively correlated with the final exam and overall unit grade, but negatively correlated with laboratory attendance rates (r₆₀ = 0.284, P < 0.05). This differed somewhat from the results of Brown et al. (2015), who found a positive correlation between the affective component and final exam score for both “all students” and “low achieving students,” but not for “high achieving students.”

The EFA showed that 34.2% of the variability in overall grades could be predicted from student responses to the ASCI instrument (results not shown). Upon closer examination of the items important to the model, items 2 (Worthless – Beneficial) and 15 (Worthwhile – Useless) were found to be the most important contributors, and the only statistically significant variables. This suggests that students’ opinion of the worth of chemistry may be significantly correlated with their performance in chemistry-related units.

Furthermore, correlation coefficients between the individual ASCI items and student achievement data revealed that item 2 (Worthless – Beneficial) had the strongest correlation coefficients with student scores on the final exam and their overall grades Table 4. Several other significant items were related to the understandability of chemistry (e.g., items 5, 9, 14) and student enjoyment (e.g., items 7, 8, 12).